JP4491273B2 - Storage system, file access control program, and file access control method - Google Patents

Description

  The present invention relates to a storage system used in an information processing system and the like, and more particularly to file access control in a file system having a plurality of storage devices or a plurality of lower-level file systems.

  In a file system that manages a large number of storage devices, it is necessary to improve both the parallelism of the plurality of storage devices and the performance of each storage device in order to exhibit high overall input / output performance. For example, when accessing a single huge file, rather than allocating and accessing the entire file to one device, the file is divided into a plurality of blocks, and each divided block is divided into a plurality of storage devices. When the plurality of storage devices are accessed in parallel, the degree of parallelism of the devices can be utilized, so that the file access performance is improved.

  However, when the number of storage devices is very large, if one file is divided into more storage devices than necessary, the performance efficiency per device deteriorates. This is because when multiple parts of the same file are assigned to the same device, a mechanism such as batch writing or prefetching of the storage device functions to improve the performance, but if the number of file divisions increases, the same device This is because the number of blocks allocated to the memory device decreases, and the storage device bulk writing and prefetching mechanisms do not function sufficiently. Further, when accessing a plurality of files at the same time, blocks of different files are allocated to the same storage device, so a seek occurs in the storage device, and the performance efficiency of the storage device further decreases.

  To solve this trade-off between parallelism and performance efficiency, the method of fixing and limiting the stripe width of a single file in file system units, and the user specifying details of the allocation information of each file to the disk, A method for speeding up file access has been proposed. In this method, a disk to be striped is designated by the user for each file, and high-speed access is realized by dividing the file into blocks having an optimum size for each file (for example, see Patent Document 1).

There has also been proposed a method in which the file system selects a file allocation method on a static basis. This method is to select a storage device in which a block size that matches the file access size is set. The unused storage area is reduced, and high-speed access is realized (see, for example, Patent Document 2). ).
Japanese Patent Laid-Open No. 11-15720 JP 2002-132548 A

  The method in which the user specifies in detail the allocation information of each file to the disk is effective in a routine business system that repeatedly uses a file once allocated. However, in a system where many general users frequently create and delete many small files, it is necessary to specify specific disk allocation information each time a user creates a file, which makes the work complicated. Actual operation is difficult.

  In the method in which the file system selects a file allocation method on a static basis, a storage device to which a file is allocated is determined based on a fixed condition such as a file access size. Therefore, neither the selection condition nor the device configuration can be changed. . For this reason, the file size and the number of files that can be accessed simultaneously are limited.

The present invention relates to a storage system comprising a node connected to be communicable with each other via a network, and a storage device connected to the node , wherein the node is a file access to be started by a user of the node A file attribute detection unit that detects a plurality of file attributes and a file access rule that is a type of access method that can be selected to perform file access using the storage device in association with each rule identifier An access rule storage unit; a file access rule selection policy storage unit that sets a file access rule selection policy that defines a rule identifier of a file access rule to be selected for each of a plurality of file attributes that may appear; and the detected the file a corresponding to a plurality of file attributes Possess the file access control unit for controlling a file access to the storage device according to Seth rules, and each of the plurality of file attributes which may be the appearance is classified into one of a static file attributes or dynamic file attributes A plurality of files detected by the file attribute detection unit are set in the file access rule selection policy storage unit as attribute values related to a predetermined attribute item, each attribute item is given a priority order, and the file access control unit For each attribute, the rule identifier of the file access rule to be selected is extracted by referring to the file access rule selection policy storage unit, and the rule identifier corresponding to the file access rule that contradicts the extracted rule identifier. If a set is included, the highest priority is given to the set. A rule identifier corresponding to an attribute value related to a sex item is selected, and one or more file access rules corresponding to one or more rule identifiers acquired by the extraction and selection are read out as a composite file access rule, and the composite file access rule According to the above, file access to the storage device is controlled .

  According to the present invention, an optimal allocation method for a file and an optimal access method for a file are automatically selected according to a policy designated by the user. Therefore, high-speed file access can be provided to the user without requiring complicated work.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of a computer system according to the embodiment of this invention.

  The client nodes 1a to 1b are communicably connected to the I / O nodes 2a to 2e via the network 3. Storage devices 4a to 4l are connected to each node.

  The network 3 is, for example, a LAN (Local Area Network).

  The client nodes 1a and 1b are computer devices used by system administrators or users, and include a CPU (not shown), a memory (not shown), a network interface (not shown), and the like. The memory of the client node 1a stores a user command or application 5 used by the user, an environment variable or setting file 18 in which system attributes and the like are set, and a file system 6a which is a file management program. For example, the user command or application 5 issues a file access command to the file system 6a.

  The client node 1b has the same configuration as the client node 1a and stores a file system 6b. The file systems 6a and 6b manage the storage devices connected to the nodes 1a to 1b and 2b to 2e and the sub file system 7 operating in the I / O node 2a.

  Next, the configuration of the file system 6a will be described.

  When the file attribute detection unit 500 is configured by a subprogram included in the file system 6a and receives a file access command from a user command or the application 5, the file attribute detection unit 500 detects the file attribute 600 corresponding to the file targeted by the command. To the file access control unit 100. Here, the file attribute 600 is, for example, a file name, a file size, a file owner name, etc. (see FIGS. 6 and 7).

  The file access control unit 100 is configured by a subprogram included in the file system 6 a and includes a file access rule selection control unit 200. The file access control unit 100 passes the file attribute 600 received from the file attribute detection unit 500 to the file access rule selection control unit 200.

  The file access rule selection control unit 200 includes subprograms included in the file access control unit 100, and applies the file access rule table 300 and the file access rule selection policy table 400 to the received file attribute 600. The composite file access rule 1000 is generated and returned to the file access control unit 100. The file access control unit 100 generates a structure definition 800 of the file 700, assigns a file area, and performs file input / output in accordance with the received composite file access rule 1000.

  Here, the file access rule is a rule applied when the client node 1 accesses a file, and is defined in advance by an administrator or a user. The file access rule table 300 is a table in which defined file access rules are stored (see FIGS. 3 to 5).

  The file access rule selection policy table 400 is a table that defines the correspondence between the file attribute 600 and the file access rule applied to the file 700 having that attribute (see FIGS. 8 to 10).

  The synthetic file access rule 1000 is a rule that is actually applied when the file 700 is accessed (see FIGS. 12 and 13). The file 700 of the file 700 defined by the file access rule selection policy table 400 is used. It is generated by combining file access rules corresponding to the attribute 600.

  Note that the file access rule selection control unit 200 does not apply the file access rule table 300 and the file access rule selection policy table 400, but the file access rule and the file access rule selection policy defined by the user using the environment variable or the setting file 18. May be applied.

  The I / O nodes 2a to 2e are computer devices that process data access from the client node 1 to the storage devices 4k to 4i, and have the same hardware configuration as that of the client node 1. When the client node 1 issues an input / output command to the storage device 4 connected to the I / O node 2, the command is transferred to the node connected to the target storage device 4 via the network 3. , Input / output is executed. Here, since the user does not directly use the I / O nodes 2a to 2e, the user command / application 5 may not be stored. Further, the I / O nodes 2a to 2e do not store the file system 6 because they are under the management of the file system 6 of the client node 1.

  Further, the I / O nodes 2a to 2e may include a sub file system that manages the directly connected storage devices. In the present embodiment, the I / O node 2a includes a sub file system (SUBFS) 7, and receives file access commands from the file system 6 of the client node 1 and manages file access to the storage devices 4k and 4l. .

  Each of the storage devices 4a to 4l is an area (a so-called logical unit) recognized as a single storage device by an operating system (not shown), and each is a single storage device (for example, a magnetic disk drive). Even if each is an aggregate of a plurality of storage devices (for example, a magnetic disk array such as RAID), each may be a single storage device or a part of a plurality of storage devices.

  In this embodiment, the file systems 6a and 6b directly manage the storage devices 4a to 4j and the storage devices 4k and 4l are managed via the sub file system 7. However, all the storage devices are directly managed. Also good. Further, all the I / O nodes 2 may be provided with a sub file system (not shown) similar to the sub file system 7, and all the storage devices may be managed via these sub file systems.

  Further, the file system 6 may be arranged not on the client node 1 but on the I / O node 2. In this case, the user command or application 5 of the client node 1 issues a file access command to the file system 6 on the I / O node 2 via the network 3, and the file system 6 on the I / O node 2 stores the storage device. 4 to manage file access.

  Further, the file system 6 may be distributed and arranged in the client node 1 and the I / O node 2.

  In addition to the client node 1 having the file system 6, a client device (not shown) not having the file system 6 is connected to the network 3, and the file system of the client node 1 is connected from the client device via the network 3. 6 may make a file access request. In this case, the file system 6 that has received the file access request from the client device manages file access to the storage device 4.

  Hereinafter, the file system 6a and the user command or application 5 operating on the client node 1a will be described in detail. The file system 6b on the client node 2b provides one file system function in cooperation with the file system 6a. Therefore, the user command or application 5 operates similarly on the client node 2b.

  FIG. 2 is an explanatory diagram of a procedure of file input / output processing according to the embodiment of this invention.

  The file input / output process includes steps of file access rule and file access rule selection policy definition by the administrator or user, file creation by the user command or application 5, and file input / output by the user command or application 5.

  Hereinafter, an outline of processing corresponding to each step will be described.

  The administrator or user defines a file access rule that specifies a file allocation rule or a file access method, and a file access rule selection policy for selecting a file access rule suitable for a target file from a plurality of file access rules. Then, the file access rule and the file access rule selection policy are notified to the file system using the graphical user interface (GUI), system call, environment variable or setting file (steps 203 and 204).

  When the user creates the file 700 (step 205), the file attribute detection unit 500 detects the file attribute 600 related to the file 700 (step 206) and passes it to the file access control unit 100 (step 207).

  The file access control unit 100 passes the file attribute 600 to the file access rule selection control unit 200, and the file access rule selection control unit 200 applies the file access rule table 300 and the file access rule selection policy table 400 to the file attribute 600. Then, a composite file access rule 1000 is created (step 208). Here, the user's environment variable or setting file 18 may be applied.

  The file access control unit 100 creates the file structure definition 800 and the file 700 using the composite file access rule 1000 (steps 209 and 210).

  When the user performs input / output access to the file 700 (step 211), the file attribute detection unit 500 detects the file attribute 600 related to the file 700 (step 212) and passes it to the file access control unit 100 (step 213).

  The file access control unit 100 passes the file attribute 600 to the file access rule selection control unit 200, and the file access rule selection control unit 200 applies the file access rule table 300 and the file access rule selection policy table 400 to the file attribute 600. Then, a composite file access rule 1000 is created (step 214). Here, the user's environment variable or setting file 18 may be applied.

  The file access control unit 100 performs file input / output using the composite file access rule 1000 (step 215).

  Next, the file access rule and the file access rule table 300 in the above steps will be described in detail.

  The file access rule is a rule for designating a method for assigning a file to the storage device 4 or a method for accessing the file. The file allocation method is a rule applied at the time of file creation, and is classified into, for example, storage area allocation, stripe method, block size, and the like.

  FIG. 3 is an explanatory diagram of the file access rule table 300 according to the embodiment of this invention.

  In FIG. 3, a rule type 301 displays the type of file access rule defined in the file access rule table 300. In the example of FIG. 3, the file access rules are classified into four types: “storage area allocation”, “stripe method”, “block size”, or “access method”.

  The file access rule ID 302 is an identifier attached to the file access rule defined in the file access rule table 300, and a unique value is assigned to each file access rule.

  The file access rule 303 column stores the contents of the file access rule defined in the file access rule table 300.

  Next, each file access rule 303 will be described.

  The file access rule 303 in which the rule type 301 is classified as “storage area allocation” designates to which storage device or sub file system the file 700 is allocated among the storage devices 4 a to 4 j or the sub file system 7. . Here, the storage area is a concept including the storage device 4 and the sub file system 7.

  A plurality of storage areas to which the file 700 is allocated can be designated. In the example of FIG. 3, the file access rules 303 whose file access rule IDs 302 are “M_A1”, “M_A2”, and “M_A3” are classified as “storage area allocation”.

  The file access rule “M_A1” is defined as “range = 4b-4d” (line 304). This means that the storage area of the file 700 is allocated to the storage devices 4b, 4c and 4d. The administrator or the user can specify an arbitrary storage area by changing the parameter “4b-4d”.

  The file access rule “M_A2” is defined as “width = 2” (line 305). This means that the file 700 is allocated to any two storage areas. Which two storage areas are allocated is not specified, and is left to the file system 6. The administrator or the user can specify an arbitrary number of storage areas by changing the parameter “2”.

  The file access rule “M_A3” is defined as “range = −1” (line 306). This means that the file 700 is assigned to all storage areas.

  In the file access rule table 300, conflicting rules such as file access rules “M_A1”, “M_A2”, and “M_A3” can be defined and stored.

  The administrator or user can add arbitrary rules having different parameters in addition to the above “M_A1” to “M_A3”. Further, the storage area may be designated by a method other than the above “range” and “width”.

  A file access rule 303 in which the rule type 301 is classified as “stripe method” designates a unit in which the file 700 is distributed and assigned to a designated storage area. In the example of FIG. 3, the file access rules 303 whose file access rule IDs 302 are “M_S1” and “M_S2” are classified as “stripe methods”.

  The file access rule “M_S1” is defined as “stripe = block” (line 307). This means that a block stripe is designated. That is, the file 700 is divided into blocks of a predetermined size, and the divided blocks are distributed and allocated to the storage areas specified by the file access rules 303 classified in the above “allocating storage areas”. When the block stripe is applied, a plurality of storage areas can be operated in parallel when accessing one file, so that the access performance to one file is improved.

  The file access rule “M_S2” is defined as “stripe = file” (line 308). This means specifying a file stripe. That is, when there are a plurality of files 700, the plurality of files 700 are distributed and allocated to the storage areas specified by the file access rules 303 classified in the above-mentioned “storage area allocation”. As a result, one file 700 is assigned to one storage area. For example, when there are n storage areas (n is an arbitrary integer) and n files are generated, one file 700 is allocated to each storage area. The file stripe has better access performance than the block stripe when the number of files accessed simultaneously is larger than the number of storage areas.

  The file access rule 303 in which the rule type 301 is classified as “block size” designates the block size when the block stripe is applied. In the example of FIG. 3, the file access rule 303 whose file access rule ID 302 is “M_B1” and “M_B2” is classified as “block size”.

  The file access rule “M_B1” is defined as “blocksize = 512” (line 309). This means that 512 KB is specified as the block size.

  The file access rule “M_B2” is defined as “blocksize = 2048” (line 310). This means that 2048 KB (that is, 2 MB) is specified as the block size.

  The administrator or user can add a rule specifying an arbitrary block size in addition to the above “M_B1” and “M_B2”. The administrator or the user can increase the parallelism of a plurality of storage areas and the performance efficiency of a single storage area by selecting a block size that matches the characteristics of the application that uses the file and the file size.

  The file access rule 303 in which the rule type 301 is classified as “access method” is a rule related to input / output (file access) of the file 700. In the example of FIG. 3, file access rules 303 whose file access rule IDs 302 are “M_M1”, “M_M2”, “M_M3”, and “M_M4” are classified as “access methods”.

  The file access rule “M_M1” is defined as “access = sequential” (line 311). This means that sequential access is designated as the access method. When sequential access is designated, continuous blocks of the file 700 are often recorded continuously in the storage area, so that access performance is improved by issuing input / output instructions together.

  The file access rule “M_M2” is defined as “access = random” (line 312). This means that random access is designated as the access method.

  The file access rule “M_M3” is defined as “file_cache = true” (line 313). This means that a file cache is performed when the file 700 is accessed.

  The file access rule “M_M4” is defined as “file_pre_read = true” (line 314). This means that the file is prefetched when the file 700 is accessed.

  Usually, access performance is improved by performing file cache and file prefetching. However, when accessing a huge file or accessing a finely distributed file, the effect of improving the performance by the file cache or file prefetching is small. In such a case, the amount of processing can be reduced and the deterioration of access performance can be prevented by not performing file cache or file prefetching processing.

  The administrator or the user can add an arbitrary rule regarding file access in addition to the above “M_M1” to “M_M4”. An administrator or a user can improve access performance by selecting an appropriate rule regarding file access at the time of file access.

  Furthermore, the administrator or the user may add a file access rule 303 classified into other than “allocation of storage area”, “stripe method”, “block size”, or “access method” to the file access rule table 300. it can.

  As described above, the file access rule table 300 stores a plurality of combinations of the file access rule ID 302 and the file access rule 303.

  FIG. 4 is a flowchart of a procedure for setting the file access rule 303 in the file access rule table 300 by the administrator or the user in the embodiment of the present invention.

  An administrator or a user sets a file access rule 303 in the file access rule table 300 by using a GUI, a system call, an environment variable, or a setting file.

  When the administrator or the user starts the setting process (401), the content of the file access rule 303 to be set is determined, and the content is transferred to the file access rule selection control unit 200 by GUI, system call, environment variable or setting file. Notification is made (402).

  Next, the file access rule selection control unit 200 registers the notified contents in the file access rule table 300 (403) and ends the processing (404).

  FIG. 5 is an explanatory diagram illustrating an example of a GUI in which the administrator or user sets the file access rule 303 in the file access rule table 300 according to the embodiment of this invention.

  An administrator or a user can operate the client node 1 to set the file access rule 303 in the file access rule table 300. At the time of this setting, the window shown in FIG. 5 is displayed on the output device (not shown) of the client node 1.

  In this window, a file access rule ID input unit 501, a file access rule input unit 502, a registration button 503, and a registered access rule display unit 504 are displayed. The administrator or the user inputs the file access rule ID 302 to be assigned to the file access rule 303 to be set in the file access rule ID input unit 501, and inputs the file access rule 303 to be set in the file access rule input unit 502.

  Next, when the registration button 503 is operated (for example, when the mouse cursor is placed and clicked), the input ID and rule set are registered in the file access rule table 300. The registered file access rule ID 302 and file access rule 303 are displayed in a list on the registered access rule display unit 504. The example of FIG. 5 displays the case where the file access rule ID 302 and the file access rule 303 of FIG. 3 are registered.

  Next, the file attribute 600 will be described.

  FIG. 6 is an explanatory diagram of the contents of the file attribute 600 according to the embodiment of this invention.

  In FIG. 6, the attribute type 603 displays the type of the file attribute 600 detected from the file 700 by the file attribute detection unit 500. In the example of FIG. 6, the file attribute 600 is classified into two types, “static file attribute” 601 and “dynamic file attribute” 602. Here, the static file attribute 601 is an attribute determined when a file is created, and corresponds to, for example, a file name and a file size. The dynamic file attribute 602 is an attribute determined for each file access, and corresponds to, for example, a program name for accessing a file, a user name for accessing a file, and the like.

  The file attribute item ID 604 is an identifier assigned to the file attribute item, and a unique value is assigned to each item.

  The attribute priority 605 is a value representing the priority given to the file attribute item. In general, one file has a plurality of file attributes 600. When the conflicting file access rules 303 correspond to the respective file attributes 600, the rule corresponding to the file attribute item having a high priority is selected and applied. In the example of FIG. 6, the priority is higher as the value of the attribute priority 605 is larger. The correspondence between the file attribute 600 and the file access rule 303 will be described in detail with reference to FIG. 8, and the selection of the file access rule 303 using the attribute priority 605 will be described in detail with reference to FIG.

  A file attribute item 606 indicates a file attribute item corresponding to the file attribute item ID 604. An attribute corresponding to the file attribute item 606 is a file attribute 600.

  In the example of FIG. 6, the file attribute item 606 corresponding to the file attribute item ID “AT1” is “directory name to which the file belongs”, the attribute priority 605 is “1”, and the attribute type 603 is “static”. “File attribute” (line 607). Here, for example, when the file is under the root directory, the file attribute 600 corresponding to the file attribute item “directory name to which the file belongs” is “/” (that is, the root directory).

  Similarly, the file attribute item ID “AT2” corresponds to the file attribute item “file name” and the attribute priority “2”, and is classified as “static file attribute” (line 608).

  The file attribute item ID “AT3” corresponds to the file attribute item “file size” and the attribute priority “3”, and is classified as “static file attribute” (line 609).

  The file attribute item ID “AT4” corresponds to the file attribute item “file owner name” (that is, the name of the file owner) and the attribute priority “4”, and is classified as “static file attribute” (line 610).

  The file attribute item ID “AT5” corresponds to the file attribute item “file owner ID” (that is, an identifier uniquely assigned to the owner of the file) and the attribute priority “5”, and is changed to “static file attribute”. Classified (line 611).

  Further, as a file attribute item classified as “static file attribute”, there may be “file creation user name” (that is, the name of the user who created the file).

  The file attribute item ID “AT6” corresponds to the file attribute item “file access program name” and the attribute priority “6”, and is classified as “dynamic file attribute” (line 612).

  The file attribute item ID “AT7” corresponds to the file attribute item “file access process ID” and the attribute priority “7”, and is classified as “dynamic file attribute” (line 613).

  The file attribute item ID “AT8” corresponds to the file attribute item “file access user name” (that is, the name of the user accessing the file) and the attribute priority “8”, and is classified as “dynamic file attribute”. (Line 614).

  The file attribute item ID “AT9” corresponds to the file attribute item “file access user ID” (that is, an identifier uniquely given to the user accessing the file) and the attribute priority “9”. (Line 615).

  The file attribute item ID “AT10” corresponds to the file attribute item “file access unit” and the attribute priority “10”, and is classified as “dynamic file attribute” (line 616).

  The file attribute detection unit 500 detects the file attribute for each file attribute item 606. The administrator or user can newly add a file attribute item 606 (not shown in FIG. 6) and attach a file attribute item ID 604 and an attribute priority 605.

  FIG. 7 is an explanatory diagram of an example of the file attribute 600 detected by the file attribute detection unit 500 according to the embodiment of this invention.

  In the example of FIG. 7, the program “fft.exe” is created in the directory “/ fast_tmp”, and the attribute of the file “fft.dat” owned by “yamada” is shown.

  The file attribute item ID 701 indicates a file attribute item ID 604 corresponding to the file attribute item 606 of the file “fft.dat” detected by the file attribute detection unit 500. The file attribute item 606 corresponding to each file attribute item ID 604 is shown in parentheses.

  The file attribute value 702 indicates the value of the file attribute 600 of the file “fft.dat” detected by the file attribute detection unit 500. In the example of FIG. 7, the file attribute value “/ (root directory)” (line 703) and “fast_tmp” (line 704) are detected as the file attribute item “directory name to which the file belongs”. Further, the file attribute value “yamada” is detected as the file attribute item “file owner name”, and the file attribute value “fft.exe” is detected as the file attribute item “file access program name”. A set of each file attribute value 702 and the corresponding file attribute item ID 701 is the file attribute 600 detected for the file “fft.dat”.

  In FIG. 7, for the sake of explanation, an example in which the file attribute detection unit 500 detects the file attribute value 702 for only the file attribute items “directory name to which the file belongs”, “file owner name”, and “file access program name”. However, the file attribute value 702 may be detected for other file attribute items (for example, “file name”).

  Next, the file access rule selection policy and the file access rule selection policy table 400 will be described.

  FIG. 8 is an explanatory diagram of the file access rule selection policy table 400 according to the embodiment of this invention.

  The file access rule selection policy is a means for defining the correspondence between the file attribute 600 and the file access rule. The file access rule selection policy table 400 stores the contents of the file access rule selection policy for each file attribute 600.

  In FIG. 8, a file attribute item ID 801 is an identifier of an item of the file attribute 600 whose correspondence with the file access rule is defined by the file access rule selection policy table 400. The file attribute item ID 604 of FIG. 6 and the file attribute of FIG. This corresponds to the attribute item ID 701. The file attribute item 606 corresponding to each file attribute item ID 604 is shown in parentheses.

  The file attribute value 802 is a value of the file attribute 600 whose correspondence with the file access rule is defined by the file access rule selection policy table 400, and corresponds to the file attribute value 702 of FIG.

  The file access rule ID 803 is a file access rule corresponding to the file attribute 600 defined by the file access rule selection policy table 400. The file access rule ID 803 is expressed by a combination of one or more of the file access rule IDs 302 in FIG.

  A set of the file attribute item ID 801, the corresponding file attribute value 802 and the file access rule ID 803 is a file access rule selection policy defined by the file access rule selection policy table 400.

  In the example of FIG. 8, the file attribute value “/” of the file attribute item ID “AT1 (directory name to which the file belongs)” includes file access rule IDs “M_A3”, “M_S2”, “M_B1”, “M_M2”, “M_M3” and “M_M4” correspond (line 804). In the file 700 belonging to the directory “/ (root directory)”, a file stripe specifying all storage areas is specified as the stripe method, 512 KB is specified as the block size, and random access is specified as the access method. This means that execution of file cache and file prefetching is specified at the time of access (see FIG. 3).

  Similarly, 2 MB is specified as the block size for the file 700 belonging to the directory “/ home” (line 805). In the file 700 belonging to the directory “/ fast_tmp”, a block stripe designating the storage devices 4b, 4c and 4d is designated as the stripe method, 2 MB is designated as the block size, sequential access is designated as the access method, and at the time of access Execution of file prefetching is designated (line 806). In the file 700 whose file owner name is “yamada”, two arbitrary storage areas are designated as allocation destinations (line 807). When the file access program name is “fft.exe”, sequential access is designated as the file access method (line 808). When the file access unit is 2048 KB to 8192 KB, 2 MB is designated as the block size (line 809).

  As shown in FIG. 8, the file access rule selection policy table 400 can store a plurality of file access rule selection policies.

  FIG. 9 is a flowchart of a procedure for setting the file access rule selection policy in the file access rule selection policy table 400 by the administrator or the user in the embodiment of the present invention.

  An administrator or a user sets an access rule selection policy in the file access rule selection policy table 400 using a GUI, a system call, an environment variable, or a setting file.

  When the administrator or the user starts the setting process (901), the content of the file access rule selection policy to be set is determined, and the content is determined by the GUI, system call, environment variable, or setting file. (902).

  Next, the file access rule selection control unit 200 registers the notified contents in the file access rule selection policy table 400 (903), and ends the processing (904).

  FIG. 10 is an explanatory diagram of an example of a GUI in which the administrator or user sets the file access rule selection policy in the file access rule selection policy table 400 in the embodiment of the present invention.

  An administrator or a user can operate the client node 1 to set a file access rule selection policy in the file access rule selection policy table 400. In this setting, a window shown in FIG. 10 is displayed on the input / output device (not shown) of the client node 1.

  In this window, a file attribute item ID input unit 1001, a file attribute value input unit 1002, a file access rule ID input unit 1003, a registration button 1004, and a registered file access rule selection policy display unit 1005 are displayed.

  An administrator or a user inputs a file attribute item ID 801 to be set to the file attribute item ID input unit 1001, inputs a file attribute value 802 to be set to the file attribute value input unit 1002, and a file access rule ID input unit 1003 The file access rule ID 803 to be set is input.

  Next, when the registration button 1004 is operated (for example, when the mouse cursor is placed and clicked), the set of the file attribute item ID 801, the file attribute value 802, and the file access rule ID 803 (that is, the file access rule selection policy) is changed to a file. It is registered in the access rule selection policy table 400. The registered file access rule selection policies are displayed in a list on the registered file access rule selection policy display unit 1005. The example of FIG. 10 displays the case where the file access rule selection policy shown in FIG. 8 is registered.

  According to the file access rule and the file access rule selection policy, the file access rule can be set in advance for the file attribute 600 such as the directory name to which the file belongs and the file owner name. As a result, the optimum file access rule can be applied without the user setting an allocation method or an access method for each file 700. In addition, since the administrator or user can add, delete, or change the file access rule and the file access rule selection policy, the file access rule and the file access are temporarily set when the application is executed or when the job is executed. It is possible to specify a rule selection policy and perform file access under optimum conditions.

  FIG. 11 is a flowchart illustrating a file access processing procedure executed by the file system 6 according to the embodiment of this invention.

  When the user starts the process of accessing the file 700 (including the process of creating the file 700) (1101), the file access control unit 100 in the file system 6 receives the file attribute 600 regarding the file 700 from the file attribute detection unit 500. Is received (1102). This file attribute 600 is detected by the file attribute detection unit 500.

  Next, when the file access control unit 100 passes the file attribute 600 to the file access rule selection control unit 200, the file access rule selection control unit 200 adds the file access rule table 300 and the file access rule selection policy table to the file attribute 600. 400 is applied to select a file access rule 303 conforming to the file attribute 600 (1103). Here, the user's environment variable or setting file 18 may be applied.

  Next, in step 1103, it is determined whether a plurality of file access rule selection policies are applied from the file access rule selection policy table 400 (1104). When a plurality of file access rule selection policies are applied, there is a possibility that the conflicting file access rule 303 is selected, so the attribute priority 605 is applied and the combined file access rule 1000 (ie, actually (File access rule applied to file access) is generated (1105), and the process proceeds to step 1106. The generation of the composite file access rule 1000 will be described in detail with reference to FIG.

  On the other hand, if a plurality of file access rule selection policies are not applied in step 1103, there is no possibility that the conflicting file access rule 303 is selected, so the selected file access rule 303 is combined with the composite file access rule 1000. And

  Next, it is determined whether or not the access started in step 1101 is creation of the file 700 (1106). As a result, when the access is creation of the file 700, the composite file access rule 1000 is applied to determine the storage area to be allocated to the file 700, and the file 700 is stored in the determined storage area (1107).

  On the other hand, if the access is not the creation of the file 700, the composite file access rule 1000 is applied to access the file 700 stored in the storage area (1108). These accesses will be described in detail with reference to FIG.

  FIG. 12 is an explanatory diagram of generation of the composite file access rule 1000 according to the embodiment of this invention.

  This explains the processing in step 1105 in FIG.

  In FIG. 12, as an example, the generation of the synthetic access rule 1000 of the file “fft.dat” created by the program “fft.exe” in the directory “/ fast_tmp” and owned by “yamada” is explained as in FIG. To do.

  In FIG. 12, a file attribute item ID 1202 and a file attribute value 1203 are the same as the file attribute item ID 701 and the file attribute value 702 in FIG. 7, respectively.

  The file access rule ID 1204 is selected in step 1103 of FIG. 11, and is selected with reference to each file attribute value 1203 and the file access rule selection policy table 400 (FIG. 8).

  The priority 1201 is the attribute priority 605 (FIG. 6) corresponding to the file attribute item ID 1202, and the higher the value, the higher the priority. Here, there are two items whose file attribute item ID 1202 is “AT1” (that is, the file attribute item is “directory name to which the file belongs”). As described above, when the file access rule 303 is assigned to each directory hierarchy, the priority order of items corresponding to lower directories is set higher.

  As described above, when the file access rule 303 corresponding to a plurality of file attribute values is selected, in principle, all the selected file access rules 303 are applied. However, when the conflicting file access rule 303 is included, the file access rule 303 corresponding to the file attribute value having the higher priority 1201 is applied.

  For example, in the file access rule ID 1204 of FIG. 12, “M_A3” in row 1205 (assigned to all storage areas), “M_A1” in row 1206 (assigned to storage areas 4b, 4c, and 4d), and “ M_A2 ”(assigned to any two storage areas) is contradictory. In this case, “M_A2” having the highest priority 1201 is applied.

  The composite file access rule 1000 generated in this way is shown in line 1209 in FIG. In this example, a block stripe (M_S1) of 2 MB units (M_B2) to any two storage areas (M_A2) is specified, and the file cache (M_M3) and file prefetch (M_M4) are enabled as access methods. Sequential access (M_M1) is designated.

  FIG. 13 is an explanatory diagram of an example of accessing the file 700 by applying the composite file access rule 1000 according to the embodiment of this invention.

  This is an example in which the composite file access rule 1000 in the row 1209 in FIG. 12 is applied to the file “fft.dat” in FIG. Here, the size of the file 700 (that is, the file “fft.dat”) is 20 MB.

  First, any two storage areas are selected according to “M_A2” of the composite file access rule 1000. In the example of FIG. 13, the storage devices 4c and 4d are selected.

  Next, according to “M_B2”, the file 700 is divided into 10 blocks of 2 MB.

  Next, according to “M_S1”, the ten divided blocks are alternately distributed to the storage devices 4c and 4d and allocated from the top block. As a result, the first, third, fifth, seventh and ninth blocks from the beginning are allocated to the storage device 4c, and the second, fourth, sixth, eighth and tenth blocks are allocated to the storage device 4d.

  Next, according to “M_M1”, these blocks are allocated to the continuous areas on the storage areas 4c and 4d. Further, according to “M_M3” and “M_M4”, an access to which file cache and file prefetching are applied is performed. By applying the file cache and file prefetching to the blocks allocated to the continuous area, prefetching and batch writing during file access function effectively, and access performance is improved.

  In the example of FIG. 13, since the stripe destination is limited to the two storage devices 4c and 4d, the contention probability of the storage device 4 can be lowered even when other files are accessed simultaneously.

  It is also possible to set a rule for generating another file 700 with a file stripe. In this case, a file 700 to which a file stripe is applied and a file 700 to which a block stripe is applied are generated on one file system 6. And can be accessed.

  As described above, according to the file access control method of the present invention, by setting the file access rule for the directory name or the file owner name, the user does not set the file allocation method or access rule for each file. Optimal file access rules can be applied. As a result, high-speed file access can be realized without impairing user convenience.

  In addition, since the administrator or user can add, delete, or change the file access rule and the file access rule selection policy, the file access rule and the file access rule selection policy can be temporarily set when an application or job is executed. It is possible to specify and perform optimum access.

  In addition, file stripe files and block stripe files can be generated and accessed on a single file system, increasing the flexibility of system operation.

  Note that the configuration of the file system 6 of the present embodiment is an example, and the present invention can be applied to various file systems different from the present embodiment. The present invention does not limit the hardware configuration such as the number of nodes, the type of network, the type of storage device, and the number of storage devices. The gist of the present invention is to select an optimal file access method using a file access rule selection policy set by the user without the user specifying a file access method for each file.

  The present invention can be applied to a file system that manages a plurality of storage devices, and contributes to improving file access performance and user convenience.

It is a block diagram which shows the structure of the computer system of embodiment of this invention. It is explanatory drawing of the procedure of the file input / output processing of embodiment of this invention. It is explanatory drawing of the file access rule table of embodiment of this invention. It is a flowchart of the procedure which sets a file access rule in the file access rule table in the embodiment of the present invention. It is explanatory drawing of the example of GUI which sets a file access rule in the file access rule table in embodiment of this invention. It is explanatory drawing of the content of the file attribute of embodiment of this invention. It is explanatory drawing of the example of the file attribute detected by the file attribute detection part of embodiment of this invention. It is explanatory drawing of the file access rule selection policy table of embodiment of this invention. It is a flowchart of the procedure which sets a file access rule selection policy in the file access rule selection policy table in the embodiment of the present invention. It is explanatory drawing of the example of GUI which sets a file access rule selection policy to the file access rule selection policy table in embodiment of this invention. It is a flowchart explaining the procedure of the file access process which the file system of embodiment of this invention performs. It is explanatory drawing of the production | generation of the synthetic | combination file access rule of embodiment of this invention. It is explanatory drawing of the example which applies the synthetic file access rule of embodiment of this invention, and accesses a file.

Explanation of symbols

1 client node,
2 I / O node 3 Network 4 Storage device 5 User command or application 6 File system (FS)
7 Sub file system (SUBFS)
DESCRIPTION OF SYMBOLS 100 File access control part 200 File access rule selection control part 300 File access rule table 400 File access rule selection policy table 500 File attribute detection part 600 File attribute 700 File 800 File structure definition 1000 Composite file access rule

Claims (4)

Nodes connected to each other via a network, and
In a storage system comprising a storage device connected to the node,
The node is
A file attribute detection unit for detecting a plurality of file attributes for file access to be started by a user of the node ;
A file access rule storage unit that stores a file access rule that is a type of an access method that can be selected to perform file access using the storage device in association with each rule identifier;
A file access rule selection policy storage unit that sets a file access rule selection policy that defines a rule identifier of a file access rule to be selected for each of a plurality of file attributes that may appear;
Have a, a file access control unit for controlling a file access to the storage device according to the file access rule corresponding to a plurality of file attributes that the detected,
Each of the plurality of file attributes that can appear is set in the file access rule selection policy storage unit as an attribute value related to a predetermined attribute item classified as either a static file attribute or a dynamic file attribute,
Each attribute item is given a priority,
The file access control unit
For each of a plurality of file attributes detected by the file attribute detection unit, a rule identifier of a file access rule to be selected is extracted by referring to the file access rule selection policy storage unit,
When a set of rule identifiers corresponding to a file access rule that conflicts with the extracted rule identifier is included, a rule identifier corresponding to an attribute value related to the attribute item having the highest priority in the set is selected,
Reading one or more file access rules corresponding to one or more rule identifiers obtained by the extraction and selection as a composite file access rule;
A storage system for controlling file access to the storage device according to the composite file access rule . Each of the predetermined attribute items is classified into the static file attribute, the directory name to which the file belongs, the name of the file, the file size, the name of the user who owns the file, or the name of the user who created the file, or It is one of a program name for accessing a file, a process identifier for accessing the file, a user name or identifier for accessing the file, or a unit for accessing the file, which is classified as a dynamic file attribute. The storage system according to claim 1. Nodes connected to each other via a network, and
  A storage device connected to the node, and a storage system file access control method comprising:
  The node is
  A file attribute detection unit for detecting a plurality of file attributes for file access to be started by a user of the node;
  A file access rule storage unit that stores a file access rule that is a type of an access method that can be selected to perform file access using the storage device in association with each rule identifier;
  A file access rule selection policy storage unit that sets a file access rule selection policy that defines a rule identifier of a file access rule to be selected for each of a plurality of file attributes that may appear;
  A file access control unit that controls file access to the storage device in accordance with the file access rules corresponding to the plurality of detected file attributes;
  Each of the plurality of file attributes that can appear is set in the file access rule selection policy storage unit as an attribute value related to a predetermined attribute item classified as either a static file attribute or a dynamic file attribute,
  Each attribute item is given a priority,
  The file access control method includes:
  The file access control unit extracts a rule identifier of a file access rule to be selected for each of a plurality of file attributes detected by the file attribute detection unit by referring to the file access rule selection policy storage unit. Procedure and
  When a set of rule identifiers corresponding to file access rules that conflict with the extracted rule identifier is included, the file access control unit supports attribute values for attribute items with the highest priority in the set. Selecting a rule identifier to
  The file access control unit reads one or more file access rules corresponding to one or more rule identifiers acquired by the extraction and selection as a composite file access rule;
  The file access control method, wherein the file access control unit controls file access to the storage device according to the composite file access rule. Each of the predetermined attribute items is classified into the static file attribute, the directory name to which the file belongs, the name of the file, the file size, the name of the user who owns the file, or the name of the user who created the file, or It is one of a program name for accessing a file, a process identifier for accessing the file, a user name or identifier for accessing the file, or a unit for accessing the file, which is classified as a dynamic file attribute. The file access control method according to claim 3.

Images